TW578068B - Method to set up film search database and recording medium thereof - Google Patents

Abstract

A method to set up film search database is disclosed. First, divide an image data with the text subtitle into multiple scenes. According to the video variation in multiple scenes, extract multiple typical images. Employ multiple typical images to create an image browsing search. Next follow the image characteristics in image data to retrieve text block. Perform text division with respect to the text block to generate multiple text characteristics. Compare the multiple text characteristics and the text in database to generate a text data. Using the text data to establish a text index table. At last, use image browsing search and the text index table to build up a film search database.

Description

578〇68

The invention belongs to the technical field of the film, and especially relates to the interactive teaching of language teaching. The present invention relates to the establishment of language teaching films with image browsing and retrieval and text index tables. The previous technicians are learning English under the international trend. ^ Self-study textbooks have emerged at the historic moment, in which interactive language stands for the same effect, but the current system of language self-study textbooks is the most suitable for consumption. It is not easy to use a product, and production is labor-intensive. , The interactive and incomparable Mubei in the market, the content of audio-visual teaching materials is not lively enough ^ Textbooks are limited, the price is fresh and feels boring. Because of this, users of Google Easy Money lost their invention. In view of this, it is to provide a method for retrieving a database of foreign language film videos: conversion to interactive. The main purpose of the present invention is to build the language teaching videos to achieve the above-mentioned purpose of the invention. The method of searching the database of Yili film has been achieved. = Send: the provided building has subtitles, and cuts the image data: the shirt looks like poor material. The change of its video is extracted from a plurality of representative t numbers.... Image browsing and retrieval. Then ^ 彳 use a plurality of representative images to generate a text block. For the text block-image feature capture in shirt image data. Plural character features and data = word cut = to generate plural character features. Use text data to build a text comparison to generate a text resource. Finally use the image

578068 V. Description of the invention (2)

Browsing and Searching and Text Indexing In order to make the present invention easier to understand, the following is specifically explained as follows: Table 袼, to build a movie search database. , And other purposes, features, and advantages to better understand the best embodiment of the car, and in conjunction with the attached drawings, detailed implementation, see Figure 1, Figure 5 to build a video retrieval database. The operation flow of the embodiment of the present invention will be described in the following operation flow. First, in the embodiment of the present invention, as in step S10, the eight red w + videos with Chinese subtitles are analyzed. For example,- The foreign language is cut into a plurality of scenes. By electricity, step s12, the movie is cut into different scenes according to the timing of the scene, first: two-moving and fast split-lens test, step SU, extracting the Λ table and cutting ^ / into a plurality of scenes. Capture of day-to-day changes of the frame '= 2: Browse and retrieve according to the view ^ t- ^^^ # (key frame) in each scene, because one segment is more than == 2 Step S16, establish the image scene , Because a representative image represents one = shirt, there may be thousands of images, and it can be completed quickly: the heart uses a plurality of representative images to see the action of loose cords through the time of these representative images. The video clip to browse. And content ', you can quickly find step S22, extract text blocks, and strengthen the action, that is, to strengthen the edge of the video, a "data t video detection (edge detect ion), to capture the receiver by the edge Detection area, due to the image in the video

578〇68 V. Description of the invention (3) The edge of the part has a large radian and is irregular. The edges of the caption part of the text in the video are mostly straight lines. Based on this image feature, the text area in the picture is extracted. Step S24 'Broadcast the text area, cut the text block based on the length, width, and ^ characters of the text block, and cut the text into multiple individual line densities and structures, and divide the color in the text block into black and white. The word block 'uses binarization to separate individual text mostly in very complicated pictures' color. Because the text in general movies uses individual text blocks to be converted to white, it is necessary to remove complex backgrounds to generate text features. The black text on the background separates the text from the background.

In step S 2 6 ′, the text features are filially matched to generate a text data, and the characters in the text and feature database are compared, and the text features are distinguished from the text characteristics and the text feature data is a text data.硪, find similar text, generate step S 28, use text data, use text data to correspond to the video, (create a text index table to search for the video clip to be played 糸 'to create a text index table, use Step S32 'Create a dictionary resource dictionary database. &Quot;' Use text data to create a

Step S30: Establish a video retrieval database, which is based on the analysis of the video and audio files, in the video retrieval inspection and greed database generated by the video retrieval. Double ', and text index table, please refer to Figure 2 to create a shadow. Figure 2 is a schematic diagram of the _ film search database.' ',' ,, and the execution of the shadow search database in the embodiment of the present invention 4 〇 Image review

578068 V. Description of the invention (4) Search 50. The user clicks on the image to browse and place the video clip where the representative image image is located. ^ List the image image, and dial the text index table 60 in the search database 40. = User can also Reliance-Retrieval Shelves, Playback and Retrieval Columns: Give way to the index. In addition, the user uses the keyboard to input the video clip represented by two pounds to the video clip to be released. Take Fang Shou as the retrieval information. At the same time, when playing the movie, use the lever _ curtain device 90 to cut the day and time into two windows! : Film ―: the library function, the movie will be played in honor. Among them, the video playback area 75 == ㈣ is used to stop the playback when dialing, redial the video, and fast forward. The function of mask =: = film. The subtitles in the film can be controlled by switching the subtitle mask two:]; that is: cover the video to cover or display the subtitles, and cover 6 and 5 JU in another window for optional 80. Display the subtitles in the played video, the word i 1. The sub-screen display does not analyze the material generated by the subtitles in the video =: = The text in the sub-text is divided into text data, and the text is displayed through a dictionary database 70; ", part of speech, film The language-related information is displayed in the subtitles = scene picture display area 85, which displays the field corresponding to the Chinese characters in the subtitle display area 80. The video retrieval is provided by the present invention, which is convenient and efficient for users to convert the general external Second, = f, language teaching materials, indexed according to the subtitles in the film = the best video browsing control method and convenient language teaching tools 〃, although the invention has been better. The embodiment disclosed: on: = it is not used for

I 0213-9045TWf (Nl); STLC-01-D9035; FRANKLIN.ptd Page 7 578068

0213-9045TWf (Nl); STLC-01-D9035; FRANKLIN.ptd Page 8 578068 Brief description of the diagram Figure 1 is a schematic diagram showing the operation flowchart of the method for establishing a movie search database according to the embodiment of the present invention; Figure 2 It is a schematic diagram showing a movie retrieval database performed in the embodiment of the present invention. Symbol description S1 0 ～ analyze image data; S1 2 ～ cut movie into multiple scenes; S1 4 ～ retrieve representative images; 51 6 ～ establish image browsing and retrieval; S 2 2 ～ retrieve text blocks; S 2 4 ～ Cut text blocks into text; 52 6 ~ Compare text features with text in the database to generate a text data; S28 ~ Use text data to create a text index table; S30 ~ Create a video retrieval database; 4 0 ~ Retrieve Data library; 50 to image browsing and retrieval; 60 to text index table; 6 to 5 subtitle mask; 67 to function column; 70 to dictionary database 7 5 to movie playback area 8 to 0 to subtitle display area 8 to 5 Picture display area

0213-9045TWf (N 1); STLC-01-D9035; FRANKLIN, ptd p. 9 578068 Schematic description 9 0 ~ screen device nil p. 10 0213-9 (H5TWf (Nl); STLC-01-D9035; FRANKLIN .ptd

Claims (1)

578068 6. Scope of patent application1. A method for creating a movie image data and searching the database from a film, including the following steps: It has subtitles, and cuts the above-mentioned image data into plural scenes; it is extracted according to the changes of the visual scene in the plural scenes A plurality of representative images; the above-mentioned text information will be used on the basis of the above; the above-mentioned video retrieval resources will be used on the above-mentioned 2. the method of Rushen, the third of which is a representational image. 4. If you apply the method, place it in place. 5. If you apply the method, you can get the information from the plural representation image text block. The plural text representation text image library. Please use the above words in the patent model to create an image browsing search based on the daytime image; extract the text blocks from the image features in the material; cut the text to generate the plural text features; compare the features with the text in the database , Generate a material, create a text index form; and browse the search and the above text index form, and create a video retrieval database as described in item 1 in Chinese and English subtitles. The video with large changes in the establishment of the video retrieval database described in item 1 is generated by the above-mentioned plural patent application patents, and the use of point-and-click search is recommended. The establishment of the video retrieval database described in item 1 of the patent scope selects the above text index A video clip represented by any of the search fields in the table. In the establishment of the video retrieval database described in item 1, a keyboard input device is used to input text data as a corresponding video clip. 0213-9045TWf (N 1); STLC-01-D9035; FRANKLIN, ptd Page 11 578068 6. Application for Patent Scope 6 · The method for establishing a movie search database as described in Item 1 of the Patent Application Scope, where For any representative image in the image index table, play the scene where the representative image is located. 7. The method for establishing a movie search database as described in item 1 of the scope of patent application, wherein the above text block is covered, and the above text subtitle is selected to be covered or displayed. 8. The method 5 of establishing a movie search database as described in item 1 of the scope of patent application, wherein the above-mentioned text is used to build a dictionary corruption database. 9. A recording medium recording a program for causing a computer to execute a method for establishing a movie retrieval database, comprising the following steps: an image data having subtitles, and the image data is cut into a plurality of scenes; Depending on the change, extract a plurality of representative images; use the plurality of representative images to establish an image browsing search; extract text blocks based on the image features in the image data; cut the text blocks to generate a plurality of text features; The plurality of text features are compared with the text in the database to generate a text data; the text data is used to create a text index table; and the image browsing search and the text index table are used to create a video search database. 10. The recording medium according to item 9 of the scope of patent application, wherein the above subtitles are Chinese and English subtitles. 0213-9045TWf (Nl); STLC-01-D9035; FRANKLIN.ptd Page 12 VI. Reading patent fungus ------------ day change 2 Fan fungus item θ Record the green media, which is based on the number of representative images on the basis of 12. Ru Shen H Hongzhi. Click i escape text; „The recording medium described in item 9 of dried bacteria, which uses the representative of the shadow t in 4 grids-check the purple bar rice, put the above search bar-key public: entry application: set patent The media recorded in item 9 of Erwei, which uses movie clips. Rotating text data as search data to obtain corresponding points, Λ4 ·, Λ, the recording medium described in item 9 of the scope of patent application, which is used. The scene where any of the representative shadowy images in。 is located. 上述 The above representative · + As described in the recording medium described in item 9 of the patent scope, Fu Zhongli m covers the above text block and chooses to cover the above text subtitles Or display: Use 1 6. If the above-mentioned patent data is recorded in the scope of the patent application, a dictionary dictionary database is established, which uses page 13 0213-9045TWf (Nl); STLC-01-D9035; FRANKLIN.ptd